Cost minimization analysis of capecitabine versus 5-fluorouracil-based treatment for gastric cancer patients in Hong Kong.

BACKGROUND: EOX (epirubicin, oxaliplatin, Xeloda; capecitabine) and FOLFOX4 (5-fluorouracil (5-FU), leucovorin, oxaliplatin) are the common chemotherapy regimens used in the treatment of advanced gastric cancer (aGC) in Hong Kong. This study aimed to compare the costs of these therapies for aGC patients from both the healthcare and societal perspectives. It should be noted that, while FOLFOX4 is routinely administered in an outpatient setting in North America and Europe, inpatient setting is adopted in Hong Kong instead, incurring hospitalization cost as a result. METHODS: Fifty-eight patients were identified from the electronic records in two public tertiary hospitals, with 45 and 13 receiving EOX and FOLFOX4 regimens, respectively. Healthcare cost was direct medical costs including drugs, clinic follow-up, hospitalization, diagnostic laboratories, and radiographs. Societal cost refers to indirect costs such as patient time and travel costs. Cost items were further classified as "expected" or "unexpected". All cost data was expressed in US dollars. RESULTS: Patients in the EOX and FOLFOX4 arm received an average of 5.3 and 7.8 cycles of treatment, respectively. The capecitabine-based regimen group had a higher expected medication cost per cycle when compared to the 5-FU-based treatment group (US$290.3 vs US$66.9, p < .001), but lower expected hospitalization costs (US$76.9 vs US$1,269.2, p < .001). The total healthcare cost and total societal cost per patient was reduced by 67.2% (US$5,691.9 vs US$17,357.4, p < .001) and 25.3% (US$3,090.5 vs US$4,135.1, p = .001), respectively, in the capecitabine-based regimen group. Sensitivity analyses based on full cycle regimen costs and net capecitabine or 5-FU/leucovorin costs still showed EOX to be less costly than FOLFOX4. CONCLUSION: The capecitabine-based regimen, EOX, was found to generate significant cost saving from both the healthcare and societal perspectives in regions in which FOLFOX4 is given in an inpatient setting.

A 12-year review on the aetiology and surgical outcomes of paediatric rhegmatogenous retinal detachments in Hong Kong.

PURPOSE: To evaluate the aetiology, clinical features, and surgical outcomes of paediatric rhegmatogenous retinal detachments (RRD) in Hong Kong.Patients and methodsThis is a retrospective consecutive case series of all patients aged 18 or under who underwent primary retinal detachment repair in the Hong Kong Eye Hospital from January 2000 to December 2012.ResultsForty-nine eyes of 47 patients were included. The mean age was 14, and the mean follow-up duration was 6.2 years. The most common aetiology for RRD was idiopathic (28.6%), followed by high myopia (24.5%), atopic dermatitis (AD) (18.4%), congenital and developmental abnormalities (16.3%), trauma (8.2%), and intraocular inflammation (4.1%). The mean preoperative visual acuity was LogMAR 1.0±0.8 (Snellen equivalent 6/60). The primary anatomical success rate in this series was 65.3%, and the final anatomical success rate was 85.7%. The mean postoperative visual acuity was LogMAR 0.9±1.2 (Snellen equivalent 6/48). Patients with congenital and developmental abnormalities or AD had worse anatomical and functional outcomes than patients who had no predisposing factor or high myopia. CONCLUSIONS: The primary and overall anatomical success rates in our series were comparable with existing literature. High myopia is the most commonly identifiable risk factor in Hong Kong and AD is associated with a higher re-detachment rates and a poor visual outcome.

The preparedness of Hong Kong medical students towards advance directives and end-of-life issues.

OBJECTIVES: To explore the preparedness of medical students towards advance directives and related end-of-life issues, and to examine background factors such as knowledge, attitudes, and experience concerning advance directives and related end-of-life issues. METHODS: In 2007, 448 medical students in years 3 to 5 were surveyed at the Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong. Their knowledge, attitudes, and experiences of advance directives and related end-of-life issues, and their self-perceived preparedness to discuss these matters with patients were evaluated. RESULTS: A total of 220 (49%) of the eligible students responded, of whom 79% supported the use of advance directives. Only 65 (30%) students were certain of what advance directives meant and 198 (90%) students felt that their knowledge of advance directives was inadequate. Also, 197 (90%) students felt unprepared about advance directives and end-of-life issues. Factors associated with positive attitude towards advance directives included religion and knowledge. No factors were found to be associated with self-perceived preparedness towards advance directives or end-of-life issues. CONCLUSIONS: Most of the medical students surveyed demonstrated a positive attitude towards advance directives and recognised the importance of advance directives. However, they felt that they were unprepared and lacking in knowledge and experience of advance directives and end-of-life issues. Wider range and more depth of education is needed to better equip medical students for future practice.

Constructive algorithms for structure learning in feedforward neural networks for regression problems.

In this survey paper, we review the constructive algorithms for structure learning in feedforward neural networks for regression problems. The basic idea is to start with a small network, then add hidden units and weights incrementally until a satisfactory solution is found. By formulating the whole problem as a state-space search, we first describe the general issues in constructive algorithms, with special emphasis on the search strategy. A taxonomy, based on the differences in the state transition mapping, the training algorithm, and the network architecture, is then presented.

Objective functions for training new hidden units in constructive neural networks.

In this paper, we study a number of objective functions for training new hidden units in constructive algorithms for multilayer feedforward networks. The aim is to derive a class of objective functions the computation of which and the corresponding weight updates can be done in O(N) time, where N is the number of training patterns. Moreover, even though input weight freezing is applied during the process for computational efficiency, the convergence property of the constructive algorithms using these objective functions is still preserved. We also propose a few computational tricks that can be used to improve the optimization of the objective functions under practical situations. Their relative performance in a set of two-dimensional regression problems is also discussed.

Use of bias term in projection pursuit learning improves approximation and convergence properties.

In a regression problem, one is given a multidimensional random vector X, the components of which are called predictor variables, and a random variable, Y, called response. A regression surface describes a general relationship between X and Y. A nonparametric regression technique that has been successfully applied to high-dimensional data is projection pursuit regression (PPR). The regression surface is approximated by a sum of empirically determined univariate functions of linear combinations of the predictors. Projection pursuit learning (PPL) formulates PPR using a 2-layer feedforward neural network. The smoothers in PPR are nonparametric, whereas those in PPL are based on Hermite functions of some predefined highest order R. We demonstrate that PPL networks in the original form do not have the universal approximation property for any finite R, and thus cannot converge to the desired function even with an arbitrarily large number of hidden units. But, by including a bias term in each linear projection of the predictor variables, PPL networks can regain these capabilities, independent of the exact choice of R. Experimentally, it is shown in this paper that this modification increases the rate of convergence with respect to the number of hidden units, improves the generalization performance, and makes it less sensitive to the setting of R. Finally, we apply PPL to chaotic time series prediction, and obtain superior results compared with the cascade-correlation architecture.